Face Generation¶

In this project, you'll use generative adversarial networks to generate new images of faces.

Get the Data¶

You'll be using two datasets in this project:

MNIST
CelebA

Since the celebA dataset is complex and you're doing GANs in a project for the first time, we want you to test your neural network on MNIST before CelebA. Running the GANs on MNIST will allow you to see how well your model trains sooner.

If you're using FloydHub, set data_dir to "/input" and use the FloydHub data ID "R5KrjnANiKVhLWAkpXhNBe".

data_dir = './data'

# FloydHub - Use with data ID "R5KrjnANiKVhLWAkpXhNBe"
#data_dir = '/input'


"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper

helper.download_extract('mnist', data_dir)
helper.download_extract('celeba', data_dir)

Found mnist Data
Found celeba Data

Explore the Data¶

MNIST¶

As you're aware, the MNIST dataset contains images of handwritten digits. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
%matplotlib inline
import os
from glob import glob
from matplotlib import pyplot

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')

<matplotlib.image.AxesImage at 0x7fc002cbe390>

CelebA¶

The CelebFaces Attributes Dataset (CelebA) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))

<matplotlib.image.AxesImage at 0x7fc002be4f60>

Preprocess the Data¶

Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.

The MNIST images are black and white images with a single color channel while the CelebA images have 3 color channels (RGB color channel).

Build the Neural Network¶

You'll build the components necessary to build a GANs by implementing the following functions below:

model_inputs
discriminator
generator
model_loss
model_opt
train

Check the Version of TensorFlow and Access to GPU¶

This will check to make sure you have the correct version of TensorFlow and access to a GPU

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer.  You are using {}'.format(tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))

TensorFlow Version: 1.0.0
Default GPU Device: /gpu:0

Input¶

Implement the model_inputs function to create TF Placeholders for the Neural Network. It should create the following placeholders:

Real input images placeholder with rank 4 using image_width, image_height, and image_channels.
Z input placeholder with rank 2 using z_dim.
Learning rate placeholder with rank 0.

Return the placeholders in the following the tuple (tensor of real input images, tensor of z data)

import problem_unittests as tests

def model_inputs(image_width, image_height, image_channels, z_dim):
    """
    Create the model inputs
    :param image_width: The input image width
    :param image_height: The input image height
    :param image_channels: The number of image channels
    :param z_dim: The dimension of Z
    :return: Tuple of (tensor of real input images, tensor of z data, learning rate)
    """
    inputs_real = tf.placeholder(tf.float32, (None, image_width, image_height, image_channels), name='inputs_real')
    inputs_z = tf.placeholder(tf.float32, (None, z_dim), name='inputs_z')
    learning_rate = tf.placeholder(tf.float32, name = 'learning_rate')

    return inputs_real, inputs_z, learning_rate


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_inputs(model_inputs)

Tests Passed

Discriminator¶

Implement discriminator to create a discriminator neural network that discriminates on images. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "discriminator" to allow the variables to be reused. The function should return a tuple of (tensor output of the discriminator, tensor logits of the discriminator).

def discriminator(images, reuse=False, n_units=128, alpha=0.1):
    """
    Create the discriminator network
    :param images: Tensor of input image(s)
    :param reuse: Boolean if the weights should be reused
    :return: Tuple of (tensor output of the discriminator, tensor logits of the discriminator)
    """
    
    with tf.variable_scope('discriminator', reuse=reuse):
        # Input layer is 32x32x3
        x1 = tf.layers.conv2d(images, 64, 5, strides=2, padding='same', kernel_initializer=tf.contrib.layers.xavier_initializer())
        relu1 = tf.maximum(alpha * x1, x1)
        # 16x16x64
        
        x2 = tf.layers.conv2d(relu1, 128, 5, strides=2, padding='same', kernel_initializer=tf.contrib.layers.xavier_initializer())
        bn2 = tf.layers.batch_normalization(x2, training=True)
        relu2 = tf.maximum(alpha * bn2, bn2)
        # 8x8x128
        
        x3 = tf.layers.conv2d(relu2, 256, 5, strides=2, padding='same', kernel_initializer=tf.contrib.layers.xavier_initializer())
        bn3 = tf.layers.batch_normalization(x3, training=True)
        relu3 = tf.maximum(alpha * bn3, bn3)
         # 4x4x256
            
        relu3 = tf.nn.dropout(relu3, keep_prob=0.8)    

        # Flatten it
        flat = tf.reshape(relu3, (-1, 4*4*256))
        logits = tf.layers.dense(flat, 1)
        out = tf.sigmoid(logits)
        
        return out, logits

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_discriminator(discriminator, tf)

Tests Passed

Generator¶

Implement generator to generate an image using z. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "generator" to allow the variables to be reused. The function should return the generated 28 x 28 x out_channel_dim images.

def generator(z, out_channel_dim, is_train=True, n_units=128, alpha=0.1):
    """
    Create the generator network
    :param z: Input z
    :param out_channel_dim: The number of channels in the output image
    :param is_train: Boolean if generator is being used for training
    :return: The tensor output of the generator
    """    
    
    with tf.variable_scope('generator', reuse=not is_train):
               # First fully connected layer
        x1 = tf.layers.dense(z, 7*7*512)
        # Reshape it to start the convolutional stack
        x1 = tf.reshape(x1, (-1, 7, 7, 512))
        x1 = tf.layers.batch_normalization(x1, training=is_train)
        x1 = tf.maximum(alpha * x1, x1)
        # 7x7x512 now
        
        x2 = tf.layers.conv2d_transpose(x1, 256, 5, strides=2, padding='same', kernel_initializer=tf.contrib.layers.xavier_initializer())
        x2 = tf.layers.batch_normalization(x2, training=is_train)
        x2 = tf.maximum(alpha * x2, x2)
        # 14x14x256 now
        
        x3 = tf.layers.conv2d_transpose(x2, 128, 5, strides=2, padding='same', kernel_initializer=tf.contrib.layers.xavier_initializer())
        x3 = tf.layers.batch_normalization(x3, training=is_train)
        x3 = tf.maximum(alpha * x3, x3)
        # 28x28x128 now
        
        # Output layer
        logits = tf.layers.conv2d_transpose(x3, out_channel_dim, 3, strides=1, padding='same', kernel_initializer=tf.contrib.layers.xavier_initializer())
        # 28x28xoutput_dim now
        
        out = tf.tanh(logits)
        
        return out
        
        
        
        # First fully connected layer
#         x1 = tf.layers.dense(z, 7*7*128)
#         # Reshape it to start the convolutional stack
#         x1 = tf.reshape(x1, (-1, 7, 7, 128))
#         x1 = tf.layers.batch_normalization(x1, training=is_train)
#         x1 = tf.maximum(alpha * x1, x1)
#         # 7x7x512 now
        
#         x2 = tf.layers.conv2d_transpose(x1, 64, 5, strides=2, padding='same')
#         x2 = tf.layers.batch_normalization(x2, training=is_train)
#         x2 = tf.maximum(alpha * x2, x2)
#         # 14x14x256 now
        
        
#         # Output layer
#         logits = tf.layers.conv2d_transpose(x2, out_channel_dim, 5, strides=2, padding='same')
#         # 28x28x3 now
        
#         out = tf.tanh(logits)
        
#         return out

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_generator(generator, tf)

Tests Passed

Loss¶

Implement model_loss to build the GANs for training and calculate the loss. The function should return a tuple of (discriminator loss, generator loss). Use the following functions you implemented:

discriminator(images, reuse=False)
generator(z, out_channel_dim, is_train=True)

def model_loss(input_real, input_z, out_channel_dim, smooth=0.1):
    """
    Get the loss for the discriminator and generator
    :param input_real: Images from the real dataset
    :param input_z: Z input
    :param out_channel_dim: The number of channels in the output image
    :return: A tuple of (discriminator loss, generator loss)
    """
    
    g_model = generator(input_z, out_channel_dim)
    d_model_real, d_logits_real = discriminator(input_real, reuse=False)
    d_model_fake, d_logits_fake = discriminator(g_model, reuse=True)

    d_loss_real = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_real, labels=tf.ones_like(d_model_real)  * (1 - smooth)))
    d_loss_fake = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.zeros_like(d_model_fake)))
    g_loss = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.ones_like(d_model_fake)))

    d_loss = d_loss_real + d_loss_fake
    
    return d_loss, g_loss


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_loss(model_loss)

Tests Passed

Optimization¶

Implement model_opt to create the optimization operations for the GANs. Use tf.trainable_variables to get all the trainable variables. Filter the variables with names that are in the discriminator and generator scope names. The function should return a tuple of (discriminator training operation, generator training operation).

def model_opt(d_loss, g_loss, learning_rate, beta1):
    """
    Get optimization operations
    :param d_loss: Discriminator loss Tensor
    :param g_loss: Generator loss Tensor
    :param learning_rate: Learning Rate Placeholder
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :return: A tuple of (discriminator training operation, generator training operation)
    """
    # TODO: Implement Function
       # Get weights and bias to update
    t_vars = tf.trainable_variables()
    d_vars = [var for var in t_vars if var.name.startswith('discriminator')]
    g_vars = [var for var in t_vars if var.name.startswith('generator')]

    # Optimize
    with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):
        d_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(d_loss, var_list=d_vars)
        g_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(g_loss, var_list=g_vars)

    return d_train_opt, g_train_opt


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_opt(model_opt, tf)

Tests Passed

Neural Network Training¶

Show Output¶

Use this function to show the current output of the generator during training. It will help you determine how well the GANs is training.

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import numpy as np

def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
    """
    Show example output for the generator
    :param sess: TensorFlow session
    :param n_images: Number of Images to display
    :param input_z: Input Z Tensor
    :param out_channel_dim: The number of channels in the output image
    :param image_mode: The mode to use for images ("RGB" or "L")
    """
    cmap = None if image_mode == 'RGB' else 'gray'
    z_dim = input_z.get_shape().as_list()[-1]
    example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])

    samples = sess.run(
        generator(input_z, out_channel_dim, False),
        feed_dict={input_z: example_z})

    images_grid = helper.images_square_grid(samples, image_mode)
    pyplot.imshow(images_grid, cmap=cmap)
    pyplot.show()

Train¶

Implement train to build and train the GANs. Use the following functions you implemented:

model_inputs(image_width, image_height, image_channels, z_dim)
model_loss(input_real, input_z, out_channel_dim)
model_opt(d_loss, g_loss, learning_rate, beta1)

Use the show_generator_output to show generator output while you train. Running show_generator_output for every batch will drastically increase training time and increase the size of the notebook. It's recommended to print the generator output every 100 batches.

def train(epoch_count, batch_size, z_dim, learning_rate_scalar, beta1, get_batches, data_shape, data_image_mode):
    """
    Train the GAN
    :param epoch_count: Number of epochs
    :param batch_size: Batch Size
    :param z_dim: Z dimension
    :param learning_rate_scalar: Learning Rate
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :param get_batches: Function to get batches
    :param data_shape: Shape of the data
    :param data_image_mode: The image mode to use for images ("RGB" or "L")
    """
    
    show_every = 100
    print_every=5
    n_images=25
    samples, losses = [], []
    steps = 0
    
    inputs_real, inputs_z, learning_rate = model_inputs(data_shape[1], data_shape[2], data_shape[3], z_dim)
    d_loss, g_loss = model_loss(inputs_real, inputs_z, data_shape[3])
    d_train_opt, g_train_opt = model_opt(d_loss, g_loss, learning_rate, beta1)
    
    
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        for epoch_i in range(epoch_count):
            for batch_images in get_batches(batch_size):
                steps += 1
                batch_images *= 2.0

                # Sample random noise for G
                batch_z = np.random.uniform(-1, 1, size=(batch_size, z_dim))

                # Run optimizers
                _ = sess.run(d_train_opt, feed_dict={inputs_real: batch_images, inputs_z: batch_z, learning_rate: learning_rate_scalar})
                _ = sess.run(g_train_opt, feed_dict={inputs_z: batch_z, inputs_real: batch_images, learning_rate: learning_rate_scalar})

                if steps % print_every == 0:
                    # At the end of each epoch, get the losses and print them out
                    train_loss_d = d_loss.eval({inputs_z: batch_z, inputs_real: batch_images})
                    train_loss_g = g_loss.eval({inputs_z: batch_z})

                    print("Step {}/{} Epoch {}/{}...".format(steps, batch_size*epoch_count, epoch_i+1, epoch_count),
                          "Discriminator Loss: {:.4f}...".format(train_loss_d),
                          "Generator Loss: {:.4f}".format(train_loss_g))
                    # Save losses to view after training
                    losses.append((train_loss_d, train_loss_g))

                if steps % show_every == 0:
                    show_generator_output(sess, n_images, inputs_z, data_shape[3], data_image_mode)
                
    return losses, samples

MNIST¶

Test your GANs architecture on MNIST. After 2 epochs, the GANs should be able to generate images that look like handwritten digits. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

batch_size = 32
z_dim = 100
learning_rate_scalar = 0.0001
beta1 = 0.1


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 2

mnist_dataset = helper.Dataset('mnist', glob(os.path.join(data_dir, 'mnist/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate_scalar, beta1, mnist_dataset.get_batches,
          mnist_dataset.shape, mnist_dataset.image_mode)

Step 5/64 Epoch 1/2... Discriminator Loss: 0.8793... Generator Loss: 1.0287
Step 10/64 Epoch 1/2... Discriminator Loss: 1.1238... Generator Loss: 0.9448
Step 15/64 Epoch 1/2... Discriminator Loss: 0.5232... Generator Loss: 2.0496
Step 20/64 Epoch 1/2... Discriminator Loss: 2.0855... Generator Loss: 0.3717
Step 25/64 Epoch 1/2... Discriminator Loss: 0.3956... Generator Loss: 3.7120
Step 30/64 Epoch 1/2... Discriminator Loss: 0.3702... Generator Loss: 4.1916
Step 35/64 Epoch 1/2... Discriminator Loss: 0.4027... Generator Loss: 3.4078
Step 40/64 Epoch 1/2... Discriminator Loss: 0.3930... Generator Loss: 3.0579
Step 45/64 Epoch 1/2... Discriminator Loss: 1.5443... Generator Loss: 3.5912
Step 50/64 Epoch 1/2... Discriminator Loss: 3.0981... Generator Loss: 0.1091
Step 55/64 Epoch 1/2... Discriminator Loss: 1.0965... Generator Loss: 2.7762
Step 60/64 Epoch 1/2... Discriminator Loss: 1.0285... Generator Loss: 1.3587
Step 65/64 Epoch 1/2... Discriminator Loss: 1.1234... Generator Loss: 0.8492
Step 70/64 Epoch 1/2... Discriminator Loss: 2.5565... Generator Loss: 3.7000
Step 75/64 Epoch 1/2... Discriminator Loss: 2.0083... Generator Loss: 0.3074
Step 80/64 Epoch 1/2... Discriminator Loss: 1.5452... Generator Loss: 0.7421
Step 85/64 Epoch 1/2... Discriminator Loss: 1.7474... Generator Loss: 0.3588
Step 90/64 Epoch 1/2... Discriminator Loss: 1.6791... Generator Loss: 1.2858
Step 95/64 Epoch 1/2... Discriminator Loss: 1.7035... Generator Loss: 0.4276
Step 100/64 Epoch 1/2... Discriminator Loss: 1.5871... Generator Loss: 0.5098

Step 105/64 Epoch 1/2... Discriminator Loss: 1.7443... Generator Loss: 0.3961
Step 110/64 Epoch 1/2... Discriminator Loss: 1.2400... Generator Loss: 1.1781
Step 115/64 Epoch 1/2... Discriminator Loss: 1.4189... Generator Loss: 0.4708
Step 120/64 Epoch 1/2... Discriminator Loss: 1.3186... Generator Loss: 1.0331
Step 125/64 Epoch 1/2... Discriminator Loss: 1.8420... Generator Loss: 0.3450
Step 130/64 Epoch 1/2... Discriminator Loss: 1.3600... Generator Loss: 0.6282
Step 135/64 Epoch 1/2... Discriminator Loss: 1.7125... Generator Loss: 0.3841
Step 140/64 Epoch 1/2... Discriminator Loss: 1.6409... Generator Loss: 1.5601
Step 145/64 Epoch 1/2... Discriminator Loss: 1.5107... Generator Loss: 0.4745
Step 150/64 Epoch 1/2... Discriminator Loss: 1.5565... Generator Loss: 0.9424
Step 155/64 Epoch 1/2... Discriminator Loss: 1.4184... Generator Loss: 0.6204
Step 160/64 Epoch 1/2... Discriminator Loss: 1.4852... Generator Loss: 0.4895
Step 165/64 Epoch 1/2... Discriminator Loss: 1.2170... Generator Loss: 0.9922
Step 170/64 Epoch 1/2... Discriminator Loss: 1.9985... Generator Loss: 0.2888
Step 175/64 Epoch 1/2... Discriminator Loss: 1.2397... Generator Loss: 0.8586
Step 180/64 Epoch 1/2... Discriminator Loss: 1.6248... Generator Loss: 0.4599
Step 185/64 Epoch 1/2... Discriminator Loss: 1.4487... Generator Loss: 1.2531
Step 190/64 Epoch 1/2... Discriminator Loss: 1.6725... Generator Loss: 0.4203
Step 195/64 Epoch 1/2... Discriminator Loss: 1.3371... Generator Loss: 1.3874
Step 200/64 Epoch 1/2... Discriminator Loss: 1.3614... Generator Loss: 0.6071

Step 205/64 Epoch 1/2... Discriminator Loss: 1.2517... Generator Loss: 1.7255
Step 210/64 Epoch 1/2... Discriminator Loss: 1.4559... Generator Loss: 0.6279
Step 215/64 Epoch 1/2... Discriminator Loss: 1.2726... Generator Loss: 0.7146
Step 220/64 Epoch 1/2... Discriminator Loss: 1.4203... Generator Loss: 0.4348
Step 225/64 Epoch 1/2... Discriminator Loss: 1.2405... Generator Loss: 1.5437
Step 230/64 Epoch 1/2... Discriminator Loss: 1.6326... Generator Loss: 0.4486
Step 235/64 Epoch 1/2... Discriminator Loss: 1.2888... Generator Loss: 1.4671
Step 240/64 Epoch 1/2... Discriminator Loss: 1.2557... Generator Loss: 0.7371
Step 245/64 Epoch 1/2... Discriminator Loss: 1.2186... Generator Loss: 0.8031
Step 250/64 Epoch 1/2... Discriminator Loss: 1.4722... Generator Loss: 0.5228
Step 255/64 Epoch 1/2... Discriminator Loss: 1.2235... Generator Loss: 1.5255
Step 260/64 Epoch 1/2... Discriminator Loss: 1.1967... Generator Loss: 0.7869
Step 265/64 Epoch 1/2... Discriminator Loss: 1.1347... Generator Loss: 0.8224
Step 270/64 Epoch 1/2... Discriminator Loss: 1.0797... Generator Loss: 1.3323
Step 275/64 Epoch 1/2... Discriminator Loss: 1.5154... Generator Loss: 0.4727
Step 280/64 Epoch 1/2... Discriminator Loss: 1.1270... Generator Loss: 0.9434
Step 285/64 Epoch 1/2... Discriminator Loss: 1.1338... Generator Loss: 1.0064
Step 290/64 Epoch 1/2... Discriminator Loss: 1.1243... Generator Loss: 0.8779
Step 295/64 Epoch 1/2... Discriminator Loss: 1.3021... Generator Loss: 0.5949
Step 300/64 Epoch 1/2... Discriminator Loss: 1.0180... Generator Loss: 1.1372

Step 305/64 Epoch 1/2... Discriminator Loss: 1.1379... Generator Loss: 0.6809
Step 310/64 Epoch 1/2... Discriminator Loss: 1.4205... Generator Loss: 1.6501
Step 315/64 Epoch 1/2... Discriminator Loss: 1.3864... Generator Loss: 0.5036
Step 320/64 Epoch 1/2... Discriminator Loss: 1.1500... Generator Loss: 0.7810
Step 325/64 Epoch 1/2... Discriminator Loss: 1.4752... Generator Loss: 0.4589
Step 330/64 Epoch 1/2... Discriminator Loss: 1.0072... Generator Loss: 1.0315
Step 335/64 Epoch 1/2... Discriminator Loss: 1.3040... Generator Loss: 0.5964
Step 340/64 Epoch 1/2... Discriminator Loss: 1.1061... Generator Loss: 1.6187
Step 345/64 Epoch 1/2... Discriminator Loss: 1.5083... Generator Loss: 0.4920
Step 350/64 Epoch 1/2... Discriminator Loss: 1.1768... Generator Loss: 0.8856
Step 355/64 Epoch 1/2... Discriminator Loss: 1.4508... Generator Loss: 0.4840
Step 360/64 Epoch 1/2... Discriminator Loss: 1.2265... Generator Loss: 1.3446
Step 365/64 Epoch 1/2... Discriminator Loss: 1.1965... Generator Loss: 0.6657
Step 370/64 Epoch 1/2... Discriminator Loss: 1.3371... Generator Loss: 2.0352
Step 375/64 Epoch 1/2... Discriminator Loss: 0.8656... Generator Loss: 1.1388
Step 380/64 Epoch 1/2... Discriminator Loss: 1.1609... Generator Loss: 1.5039
Step 385/64 Epoch 1/2... Discriminator Loss: 1.2816... Generator Loss: 0.5963
Step 390/64 Epoch 1/2... Discriminator Loss: 1.2297... Generator Loss: 0.6304
Step 395/64 Epoch 1/2... Discriminator Loss: 1.0431... Generator Loss: 1.4775
Step 400/64 Epoch 1/2... Discriminator Loss: 1.6799... Generator Loss: 0.4042

Step 405/64 Epoch 1/2... Discriminator Loss: 0.9360... Generator Loss: 1.7299
Step 410/64 Epoch 1/2... Discriminator Loss: 1.1997... Generator Loss: 0.7970
Step 415/64 Epoch 1/2... Discriminator Loss: 1.0181... Generator Loss: 0.8220
Step 420/64 Epoch 1/2... Discriminator Loss: 0.9252... Generator Loss: 1.4309
Step 425/64 Epoch 1/2... Discriminator Loss: 0.9928... Generator Loss: 1.2196
Step 430/64 Epoch 1/2... Discriminator Loss: 1.0128... Generator Loss: 0.9903
Step 435/64 Epoch 1/2... Discriminator Loss: 1.6115... Generator Loss: 0.4321
Step 440/64 Epoch 1/2... Discriminator Loss: 0.9171... Generator Loss: 1.9239
Step 445/64 Epoch 1/2... Discriminator Loss: 1.6409... Generator Loss: 0.3840
Step 450/64 Epoch 1/2... Discriminator Loss: 1.1407... Generator Loss: 0.7595
Step 455/64 Epoch 1/2... Discriminator Loss: 0.9984... Generator Loss: 1.2045
Step 460/64 Epoch 1/2... Discriminator Loss: 1.2766... Generator Loss: 0.6475
Step 465/64 Epoch 1/2... Discriminator Loss: 1.0432... Generator Loss: 1.4546
Step 470/64 Epoch 1/2... Discriminator Loss: 1.2150... Generator Loss: 0.6260
Step 475/64 Epoch 1/2... Discriminator Loss: 1.0441... Generator Loss: 0.8459
Step 480/64 Epoch 1/2... Discriminator Loss: 0.8504... Generator Loss: 1.1083
Step 485/64 Epoch 1/2... Discriminator Loss: 1.3850... Generator Loss: 2.4001
Step 490/64 Epoch 1/2... Discriminator Loss: 1.1712... Generator Loss: 0.6915
Step 495/64 Epoch 1/2... Discriminator Loss: 0.9423... Generator Loss: 1.7992
Step 500/64 Epoch 1/2... Discriminator Loss: 1.1633... Generator Loss: 0.8166

Step 505/64 Epoch 1/2... Discriminator Loss: 1.0812... Generator Loss: 0.8339
Step 510/64 Epoch 1/2... Discriminator Loss: 1.2948... Generator Loss: 0.6626
Step 515/64 Epoch 1/2... Discriminator Loss: 0.9075... Generator Loss: 1.6719
Step 520/64 Epoch 1/2... Discriminator Loss: 0.9447... Generator Loss: 1.3472
Step 525/64 Epoch 1/2... Discriminator Loss: 1.1863... Generator Loss: 0.6578
Step 530/64 Epoch 1/2... Discriminator Loss: 0.9475... Generator Loss: 1.3732
Step 535/64 Epoch 1/2... Discriminator Loss: 1.3604... Generator Loss: 0.5542
Step 540/64 Epoch 1/2... Discriminator Loss: 1.7441... Generator Loss: 0.3613
Step 545/64 Epoch 1/2... Discriminator Loss: 1.0954... Generator Loss: 0.9209
Step 550/64 Epoch 1/2... Discriminator Loss: 1.2674... Generator Loss: 0.6493
Step 555/64 Epoch 1/2... Discriminator Loss: 0.9982... Generator Loss: 1.2344
Step 560/64 Epoch 1/2... Discriminator Loss: 1.1042... Generator Loss: 0.8333
Step 565/64 Epoch 1/2... Discriminator Loss: 1.2916... Generator Loss: 0.6292
Step 570/64 Epoch 1/2... Discriminator Loss: 1.5406... Generator Loss: 0.4243
Step 575/64 Epoch 1/2... Discriminator Loss: 1.0372... Generator Loss: 1.1380
Step 580/64 Epoch 1/2... Discriminator Loss: 1.3032... Generator Loss: 0.8472
Step 585/64 Epoch 1/2... Discriminator Loss: 1.1833... Generator Loss: 0.6158
Step 590/64 Epoch 1/2... Discriminator Loss: 1.1358... Generator Loss: 1.2651
Step 595/64 Epoch 1/2... Discriminator Loss: 1.6312... Generator Loss: 0.3947
Step 600/64 Epoch 1/2... Discriminator Loss: 1.4645... Generator Loss: 2.4862

Step 605/64 Epoch 1/2... Discriminator Loss: 1.2140... Generator Loss: 0.6971
Step 610/64 Epoch 1/2... Discriminator Loss: 0.9426... Generator Loss: 1.1304
Step 615/64 Epoch 1/2... Discriminator Loss: 0.9363... Generator Loss: 2.0882
Step 620/64 Epoch 1/2... Discriminator Loss: 1.3915... Generator Loss: 0.6333
Step 625/64 Epoch 1/2... Discriminator Loss: 1.1535... Generator Loss: 1.7346
Step 630/64 Epoch 1/2... Discriminator Loss: 1.0125... Generator Loss: 1.0170
Step 635/64 Epoch 1/2... Discriminator Loss: 2.0437... Generator Loss: 0.2666
Step 640/64 Epoch 1/2... Discriminator Loss: 1.1884... Generator Loss: 1.1043
Step 645/64 Epoch 1/2... Discriminator Loss: 1.6099... Generator Loss: 0.4431
Step 650/64 Epoch 1/2... Discriminator Loss: 1.1854... Generator Loss: 1.6569
Step 655/64 Epoch 1/2... Discriminator Loss: 1.1748... Generator Loss: 0.8416
Step 660/64 Epoch 1/2... Discriminator Loss: 1.2481... Generator Loss: 1.9887
Step 665/64 Epoch 1/2... Discriminator Loss: 1.0766... Generator Loss: 1.0867
Step 670/64 Epoch 1/2... Discriminator Loss: 1.3489... Generator Loss: 0.6389
Step 675/64 Epoch 1/2... Discriminator Loss: 1.6252... Generator Loss: 0.4476
Step 680/64 Epoch 1/2... Discriminator Loss: 1.1811... Generator Loss: 1.5677
Step 685/64 Epoch 1/2... Discriminator Loss: 1.0583... Generator Loss: 1.4644
Step 690/64 Epoch 1/2... Discriminator Loss: 1.2373... Generator Loss: 0.7103
Step 695/64 Epoch 1/2... Discriminator Loss: 1.1865... Generator Loss: 1.0505
Step 700/64 Epoch 1/2... Discriminator Loss: 1.5381... Generator Loss: 0.5331

Step 705/64 Epoch 1/2... Discriminator Loss: 1.3280... Generator Loss: 1.6009
Step 710/64 Epoch 1/2... Discriminator Loss: 1.2091... Generator Loss: 1.1162
Step 715/64 Epoch 1/2... Discriminator Loss: 1.4181... Generator Loss: 0.5472
Step 720/64 Epoch 1/2... Discriminator Loss: 1.5440... Generator Loss: 0.4589
Step 725/64 Epoch 1/2... Discriminator Loss: 1.0703... Generator Loss: 1.1656
Step 730/64 Epoch 1/2... Discriminator Loss: 1.1619... Generator Loss: 1.4277
Step 735/64 Epoch 1/2... Discriminator Loss: 1.1621... Generator Loss: 0.8208
Step 740/64 Epoch 1/2... Discriminator Loss: 1.1039... Generator Loss: 1.2466
Step 745/64 Epoch 1/2... Discriminator Loss: 1.5074... Generator Loss: 0.5218
Step 750/64 Epoch 1/2... Discriminator Loss: 1.1319... Generator Loss: 1.2207
Step 755/64 Epoch 1/2... Discriminator Loss: 1.4696... Generator Loss: 0.4940
Step 760/64 Epoch 1/2... Discriminator Loss: 1.3772... Generator Loss: 1.7139
Step 765/64 Epoch 1/2... Discriminator Loss: 1.3492... Generator Loss: 0.5445
Step 770/64 Epoch 1/2... Discriminator Loss: 1.0833... Generator Loss: 1.0956
Step 775/64 Epoch 1/2... Discriminator Loss: 1.1172... Generator Loss: 1.0658
Step 780/64 Epoch 1/2... Discriminator Loss: 1.4578... Generator Loss: 0.4380
Step 785/64 Epoch 1/2... Discriminator Loss: 1.3140... Generator Loss: 0.6919
Step 790/64 Epoch 1/2... Discriminator Loss: 1.0203... Generator Loss: 1.3838
Step 795/64 Epoch 1/2... Discriminator Loss: 1.7027... Generator Loss: 0.3711
Step 800/64 Epoch 1/2... Discriminator Loss: 1.1350... Generator Loss: 0.9754

Step 805/64 Epoch 1/2... Discriminator Loss: 1.4889... Generator Loss: 0.4508
Step 810/64 Epoch 1/2... Discriminator Loss: 1.3267... Generator Loss: 0.7234
Step 815/64 Epoch 1/2... Discriminator Loss: 1.0672... Generator Loss: 1.0824
Step 820/64 Epoch 1/2... Discriminator Loss: 1.1192... Generator Loss: 1.7321
Step 825/64 Epoch 1/2... Discriminator Loss: 1.2576... Generator Loss: 1.2173
Step 830/64 Epoch 1/2... Discriminator Loss: 1.2285... Generator Loss: 0.7141
Step 835/64 Epoch 1/2... Discriminator Loss: 1.1986... Generator Loss: 0.6764
Step 840/64 Epoch 1/2... Discriminator Loss: 1.0589... Generator Loss: 1.5237
Step 845/64 Epoch 1/2... Discriminator Loss: 1.7439... Generator Loss: 0.3126
Step 850/64 Epoch 1/2... Discriminator Loss: 1.1215... Generator Loss: 1.0394
Step 855/64 Epoch 1/2... Discriminator Loss: 1.0548... Generator Loss: 0.8717
Step 860/64 Epoch 1/2... Discriminator Loss: 1.4272... Generator Loss: 0.5228
Step 865/64 Epoch 1/2... Discriminator Loss: 1.1059... Generator Loss: 1.6100
Step 870/64 Epoch 1/2... Discriminator Loss: 1.4569... Generator Loss: 0.4749
Step 875/64 Epoch 1/2... Discriminator Loss: 1.0774... Generator Loss: 1.1985
Step 880/64 Epoch 1/2... Discriminator Loss: 1.2577... Generator Loss: 0.9116
Step 885/64 Epoch 1/2... Discriminator Loss: 1.4833... Generator Loss: 0.4855
Step 890/64 Epoch 1/2... Discriminator Loss: 1.1229... Generator Loss: 0.7724
Step 895/64 Epoch 1/2... Discriminator Loss: 1.1990... Generator Loss: 1.5310
Step 900/64 Epoch 1/2... Discriminator Loss: 1.5519... Generator Loss: 0.4992

Step 905/64 Epoch 1/2... Discriminator Loss: 1.0659... Generator Loss: 1.2291
Step 910/64 Epoch 1/2... Discriminator Loss: 1.5820... Generator Loss: 0.4849
Step 915/64 Epoch 1/2... Discriminator Loss: 1.2234... Generator Loss: 1.2296
Step 920/64 Epoch 1/2... Discriminator Loss: 1.3177... Generator Loss: 0.6531
Step 925/64 Epoch 1/2... Discriminator Loss: 1.1089... Generator Loss: 1.0191
Step 930/64 Epoch 1/2... Discriminator Loss: 1.7661... Generator Loss: 0.3760
Step 935/64 Epoch 1/2... Discriminator Loss: 1.3432... Generator Loss: 0.5800
Step 940/64 Epoch 1/2... Discriminator Loss: 1.1204... Generator Loss: 1.2636
Step 945/64 Epoch 1/2... Discriminator Loss: 1.2320... Generator Loss: 0.7354
Step 950/64 Epoch 1/2... Discriminator Loss: 1.0793... Generator Loss: 1.0090
Step 955/64 Epoch 1/2... Discriminator Loss: 1.2024... Generator Loss: 0.7562
Step 960/64 Epoch 1/2... Discriminator Loss: 1.1682... Generator Loss: 1.1788
Step 965/64 Epoch 1/2... Discriminator Loss: 1.5502... Generator Loss: 0.5914
Step 970/64 Epoch 1/2... Discriminator Loss: 1.1593... Generator Loss: 1.5755
Step 975/64 Epoch 1/2... Discriminator Loss: 1.2232... Generator Loss: 0.7000
Step 980/64 Epoch 1/2... Discriminator Loss: 1.2077... Generator Loss: 0.6577
Step 985/64 Epoch 1/2... Discriminator Loss: 1.1297... Generator Loss: 1.2397
Step 990/64 Epoch 1/2... Discriminator Loss: 1.3803... Generator Loss: 0.6436
Step 995/64 Epoch 1/2... Discriminator Loss: 1.0853... Generator Loss: 0.7909
Step 1000/64 Epoch 1/2... Discriminator Loss: 1.2098... Generator Loss: 0.9923

CelebA¶

Run your GANs on CelebA. It will take around 20 minutes on the average GPU to run one epoch. You can run the whole epoch or stop when it starts to generate realistic faces.

batch_size = 32
z_dim = 100
learning_rate = 0.0001
beta1 = 0.1


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 1

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)

Step 5/32 Epoch 1/1... Discriminator Loss: 3.4148... Generator Loss: 0.0778
Step 10/32 Epoch 1/1... Discriminator Loss: 3.4638... Generator Loss: 0.0724
Step 15/32 Epoch 1/1... Discriminator Loss: 2.4507... Generator Loss: 0.2216
Step 20/32 Epoch 1/1... Discriminator Loss: 1.8974... Generator Loss: 0.4263
Step 25/32 Epoch 1/1... Discriminator Loss: 0.9658... Generator Loss: 1.2527
Step 30/32 Epoch 1/1... Discriminator Loss: 0.7339... Generator Loss: 1.5861
Step 35/32 Epoch 1/1... Discriminator Loss: 0.8422... Generator Loss: 1.2955
Step 40/32 Epoch 1/1... Discriminator Loss: 1.1525... Generator Loss: 0.7726
Step 45/32 Epoch 1/1... Discriminator Loss: 1.1401... Generator Loss: 5.0152
Step 50/32 Epoch 1/1... Discriminator Loss: 0.8285... Generator Loss: 1.1764
Step 55/32 Epoch 1/1... Discriminator Loss: 0.7860... Generator Loss: 3.0916
Step 60/32 Epoch 1/1... Discriminator Loss: 1.7473... Generator Loss: 0.4007
Step 65/32 Epoch 1/1... Discriminator Loss: 0.6582... Generator Loss: 2.4542
Step 70/32 Epoch 1/1... Discriminator Loss: 0.9713... Generator Loss: 2.0377
Step 75/32 Epoch 1/1... Discriminator Loss: 0.7929... Generator Loss: 1.2938
Step 80/32 Epoch 1/1... Discriminator Loss: 1.7347... Generator Loss: 0.3535
Step 85/32 Epoch 1/1... Discriminator Loss: 1.1105... Generator Loss: 2.9798
Step 90/32 Epoch 1/1... Discriminator Loss: 0.7723... Generator Loss: 1.4588
Step 95/32 Epoch 1/1... Discriminator Loss: 0.7909... Generator Loss: 1.5820
Step 100/32 Epoch 1/1... Discriminator Loss: 0.7552... Generator Loss: 1.5577

Step 105/32 Epoch 1/1... Discriminator Loss: 0.5683... Generator Loss: 2.1804
Step 110/32 Epoch 1/1... Discriminator Loss: 1.3993... Generator Loss: 4.7550
Step 115/32 Epoch 1/1... Discriminator Loss: 1.6823... Generator Loss: 0.3347
Step 120/32 Epoch 1/1... Discriminator Loss: 0.8437... Generator Loss: 2.6463
Step 125/32 Epoch 1/1... Discriminator Loss: 0.7257... Generator Loss: 1.6916
Step 130/32 Epoch 1/1... Discriminator Loss: 0.6301... Generator Loss: 1.8290
Step 135/32 Epoch 1/1... Discriminator Loss: 2.7563... Generator Loss: 0.1434
Step 140/32 Epoch 1/1... Discriminator Loss: 0.5389... Generator Loss: 2.5378
Step 145/32 Epoch 1/1... Discriminator Loss: 1.5477... Generator Loss: 0.4149
Step 150/32 Epoch 1/1... Discriminator Loss: 1.0387... Generator Loss: 1.0146
Step 155/32 Epoch 1/1... Discriminator Loss: 1.4860... Generator Loss: 0.5747
Step 160/32 Epoch 1/1... Discriminator Loss: 0.8452... Generator Loss: 2.9588
Step 165/32 Epoch 1/1... Discriminator Loss: 0.7138... Generator Loss: 1.9118
Step 170/32 Epoch 1/1... Discriminator Loss: 0.7677... Generator Loss: 1.4672
Step 175/32 Epoch 1/1... Discriminator Loss: 0.7385... Generator Loss: 1.3552
Step 180/32 Epoch 1/1... Discriminator Loss: 1.0894... Generator Loss: 4.0260
Step 185/32 Epoch 1/1... Discriminator Loss: 0.7862... Generator Loss: 1.3028
Step 190/32 Epoch 1/1... Discriminator Loss: 0.6818... Generator Loss: 3.5837
Step 195/32 Epoch 1/1... Discriminator Loss: 0.5727... Generator Loss: 3.0850
Step 200/32 Epoch 1/1... Discriminator Loss: 0.7316... Generator Loss: 3.5358

Step 205/32 Epoch 1/1... Discriminator Loss: 1.1921... Generator Loss: 0.6600
Step 210/32 Epoch 1/1... Discriminator Loss: 1.0690... Generator Loss: 4.1474
Step 215/32 Epoch 1/1... Discriminator Loss: 2.0699... Generator Loss: 0.2144
Step 220/32 Epoch 1/1... Discriminator Loss: 1.1375... Generator Loss: 3.6818
Step 225/32 Epoch 1/1... Discriminator Loss: 1.9435... Generator Loss: 0.3404
Step 230/32 Epoch 1/1... Discriminator Loss: 0.7354... Generator Loss: 1.3296
Step 235/32 Epoch 1/1... Discriminator Loss: 0.6742... Generator Loss: 1.5007
Step 240/32 Epoch 1/1... Discriminator Loss: 1.3399... Generator Loss: 0.6003
Step 245/32 Epoch 1/1... Discriminator Loss: 0.8133... Generator Loss: 3.7794
Step 250/32 Epoch 1/1... Discriminator Loss: 0.6418... Generator Loss: 2.5171
Step 255/32 Epoch 1/1... Discriminator Loss: 1.4520... Generator Loss: 0.5363
Step 260/32 Epoch 1/1... Discriminator Loss: 0.5381... Generator Loss: 2.9058
Step 265/32 Epoch 1/1... Discriminator Loss: 1.2954... Generator Loss: 0.6928
Step 270/32 Epoch 1/1... Discriminator Loss: 0.6026... Generator Loss: 2.2327
Step 275/32 Epoch 1/1... Discriminator Loss: 0.6255... Generator Loss: 2.2752
Step 280/32 Epoch 1/1... Discriminator Loss: 1.6597... Generator Loss: 0.4523
Step 285/32 Epoch 1/1... Discriminator Loss: 0.5689... Generator Loss: 2.2571
Step 290/32 Epoch 1/1... Discriminator Loss: 1.7476... Generator Loss: 0.3427
Step 295/32 Epoch 1/1... Discriminator Loss: 0.6356... Generator Loss: 2.7024
Step 300/32 Epoch 1/1... Discriminator Loss: 0.7165... Generator Loss: 1.4878

Step 305/32 Epoch 1/1... Discriminator Loss: 0.8989... Generator Loss: 3.5413
Step 310/32 Epoch 1/1... Discriminator Loss: 0.8131... Generator Loss: 1.2522
Step 315/32 Epoch 1/1... Discriminator Loss: 0.5094... Generator Loss: 3.2567
Step 320/32 Epoch 1/1... Discriminator Loss: 0.8924... Generator Loss: 1.0196
Step 325/32 Epoch 1/1... Discriminator Loss: 0.5802... Generator Loss: 2.2674
Step 330/32 Epoch 1/1... Discriminator Loss: 0.8878... Generator Loss: 1.1738
Step 335/32 Epoch 1/1... Discriminator Loss: 0.6095... Generator Loss: 2.9908
Step 340/32 Epoch 1/1... Discriminator Loss: 1.1686... Generator Loss: 0.6399
Step 345/32 Epoch 1/1... Discriminator Loss: 0.5308... Generator Loss: 2.9279
Step 350/32 Epoch 1/1... Discriminator Loss: 0.8710... Generator Loss: 1.1300
Step 355/32 Epoch 1/1... Discriminator Loss: 0.5750... Generator Loss: 1.7593
Step 360/32 Epoch 1/1... Discriminator Loss: 0.5657... Generator Loss: 2.8555
Step 365/32 Epoch 1/1... Discriminator Loss: 1.2673... Generator Loss: 0.5990
Step 370/32 Epoch 1/1... Discriminator Loss: 0.7504... Generator Loss: 1.3992
Step 375/32 Epoch 1/1... Discriminator Loss: 0.5037... Generator Loss: 4.0217
Step 380/32 Epoch 1/1... Discriminator Loss: 0.9290... Generator Loss: 0.9625
Step 385/32 Epoch 1/1... Discriminator Loss: 0.9342... Generator Loss: 0.9930
Step 390/32 Epoch 1/1... Discriminator Loss: 0.5604... Generator Loss: 2.2166
Step 395/32 Epoch 1/1... Discriminator Loss: 0.5351... Generator Loss: 4.1710
Step 400/32 Epoch 1/1... Discriminator Loss: 0.5549... Generator Loss: 4.3802

Step 405/32 Epoch 1/1... Discriminator Loss: 1.0477... Generator Loss: 0.8817
Step 410/32 Epoch 1/1... Discriminator Loss: 0.5231... Generator Loss: 2.1330
Step 415/32 Epoch 1/1... Discriminator Loss: 1.2350... Generator Loss: 0.6920
Step 420/32 Epoch 1/1... Discriminator Loss: 0.4748... Generator Loss: 2.7980
Step 425/32 Epoch 1/1... Discriminator Loss: 1.9233... Generator Loss: 0.3036
Step 430/32 Epoch 1/1... Discriminator Loss: 0.9552... Generator Loss: 3.0534
Step 435/32 Epoch 1/1... Discriminator Loss: 1.2290... Generator Loss: 0.6283
Step 440/32 Epoch 1/1... Discriminator Loss: 0.5858... Generator Loss: 2.5808
Step 445/32 Epoch 1/1... Discriminator Loss: 0.8982... Generator Loss: 1.1472
Step 450/32 Epoch 1/1... Discriminator Loss: 0.8049... Generator Loss: 4.1533
Step 455/32 Epoch 1/1... Discriminator Loss: 0.8309... Generator Loss: 1.3000
Step 460/32 Epoch 1/1... Discriminator Loss: 1.6203... Generator Loss: 3.4364
Step 465/32 Epoch 1/1... Discriminator Loss: 0.4756... Generator Loss: 3.5411
Step 470/32 Epoch 1/1... Discriminator Loss: 0.7475... Generator Loss: 1.8253
Step 475/32 Epoch 1/1... Discriminator Loss: 0.6039... Generator Loss: 1.7676
Step 480/32 Epoch 1/1... Discriminator Loss: 1.0350... Generator Loss: 1.1052
Step 485/32 Epoch 1/1... Discriminator Loss: 1.0708... Generator Loss: 0.8483
Step 490/32 Epoch 1/1... Discriminator Loss: 0.9075... Generator Loss: 0.8909
Step 495/32 Epoch 1/1... Discriminator Loss: 1.7747... Generator Loss: 0.4254
Step 500/32 Epoch 1/1... Discriminator Loss: 1.3919... Generator Loss: 1.3102

Step 505/32 Epoch 1/1... Discriminator Loss: 0.5997... Generator Loss: 5.5524
Step 510/32 Epoch 1/1... Discriminator Loss: 1.6099... Generator Loss: 0.3833
Step 515/32 Epoch 1/1... Discriminator Loss: 1.0247... Generator Loss: 0.9666
Step 520/32 Epoch 1/1... Discriminator Loss: 0.5306... Generator Loss: 5.9649
Step 525/32 Epoch 1/1... Discriminator Loss: 1.3306... Generator Loss: 0.6224
Step 530/32 Epoch 1/1... Discriminator Loss: 0.7912... Generator Loss: 3.6752
Step 535/32 Epoch 1/1... Discriminator Loss: 0.6990... Generator Loss: 2.0722
Step 540/32 Epoch 1/1... Discriminator Loss: 0.6602... Generator Loss: 2.1333
Step 545/32 Epoch 1/1... Discriminator Loss: 0.7358... Generator Loss: 2.2881
Step 550/32 Epoch 1/1... Discriminator Loss: 0.9479... Generator Loss: 0.9680
Step 555/32 Epoch 1/1... Discriminator Loss: 0.8524... Generator Loss: 2.1221
Step 560/32 Epoch 1/1... Discriminator Loss: 1.7093... Generator Loss: 0.4241
Step 565/32 Epoch 1/1... Discriminator Loss: 1.2952... Generator Loss: 1.8062
Step 570/32 Epoch 1/1... Discriminator Loss: 0.5957... Generator Loss: 2.2525
Step 575/32 Epoch 1/1... Discriminator Loss: 2.8223... Generator Loss: 0.1316
Step 580/32 Epoch 1/1... Discriminator Loss: 0.7306... Generator Loss: 2.6379
Step 585/32 Epoch 1/1... Discriminator Loss: 1.7885... Generator Loss: 0.3720
Step 590/32 Epoch 1/1... Discriminator Loss: 1.4304... Generator Loss: 0.8774
Step 595/32 Epoch 1/1... Discriminator Loss: 1.0596... Generator Loss: 0.8940
Step 600/32 Epoch 1/1... Discriminator Loss: 0.7059... Generator Loss: 2.6517

Step 605/32 Epoch 1/1... Discriminator Loss: 1.2580... Generator Loss: 0.7787
Step 610/32 Epoch 1/1... Discriminator Loss: 0.8111... Generator Loss: 3.5178
Step 615/32 Epoch 1/1... Discriminator Loss: 1.3562... Generator Loss: 0.6457
Step 620/32 Epoch 1/1... Discriminator Loss: 1.3661... Generator Loss: 3.1122
Step 625/32 Epoch 1/1... Discriminator Loss: 1.6976... Generator Loss: 0.3405
Step 630/32 Epoch 1/1... Discriminator Loss: 0.8801... Generator Loss: 3.6342
Step 635/32 Epoch 1/1... Discriminator Loss: 1.7016... Generator Loss: 0.4194
Step 640/32 Epoch 1/1... Discriminator Loss: 0.8594... Generator Loss: 0.9947
Step 645/32 Epoch 1/1... Discriminator Loss: 0.8688... Generator Loss: 1.8294
Step 650/32 Epoch 1/1... Discriminator Loss: 1.4864... Generator Loss: 0.4909
Step 655/32 Epoch 1/1... Discriminator Loss: 1.1066... Generator Loss: 2.6912
Step 660/32 Epoch 1/1... Discriminator Loss: 2.0814... Generator Loss: 0.2567
Step 665/32 Epoch 1/1... Discriminator Loss: 0.8986... Generator Loss: 1.4192
Step 670/32 Epoch 1/1... Discriminator Loss: 0.7407... Generator Loss: 1.3894
Step 675/32 Epoch 1/1... Discriminator Loss: 1.0156... Generator Loss: 0.8216
Step 680/32 Epoch 1/1... Discriminator Loss: 1.2078... Generator Loss: 2.0913
Step 685/32 Epoch 1/1... Discriminator Loss: 0.5411... Generator Loss: 2.1703
Step 690/32 Epoch 1/1... Discriminator Loss: 0.9359... Generator Loss: 3.8715
Step 695/32 Epoch 1/1... Discriminator Loss: 0.8475... Generator Loss: 1.2877
Step 700/32 Epoch 1/1... Discriminator Loss: 1.0066... Generator Loss: 5.1296

Step 705/32 Epoch 1/1... Discriminator Loss: 1.7015... Generator Loss: 0.4327
Step 710/32 Epoch 1/1... Discriminator Loss: 1.0352... Generator Loss: 2.3346
Step 715/32 Epoch 1/1... Discriminator Loss: 0.5829... Generator Loss: 2.0188
Step 720/32 Epoch 1/1... Discriminator Loss: 0.6538... Generator Loss: 2.2743
Step 725/32 Epoch 1/1... Discriminator Loss: 2.4210... Generator Loss: 0.1887
Step 730/32 Epoch 1/1... Discriminator Loss: 1.0974... Generator Loss: 1.8917
Step 735/32 Epoch 1/1... Discriminator Loss: 1.1932... Generator Loss: 0.6314
Step 740/32 Epoch 1/1... Discriminator Loss: 1.7018... Generator Loss: 1.8915
Step 745/32 Epoch 1/1... Discriminator Loss: 0.7883... Generator Loss: 1.2473
Step 750/32 Epoch 1/1... Discriminator Loss: 0.5190... Generator Loss: 2.6223
Step 755/32 Epoch 1/1... Discriminator Loss: 0.6249... Generator Loss: 2.8474
Step 760/32 Epoch 1/1... Discriminator Loss: 0.5801... Generator Loss: 2.3331
Step 765/32 Epoch 1/1... Discriminator Loss: 0.6729... Generator Loss: 1.4955
Step 770/32 Epoch 1/1... Discriminator Loss: 1.1098... Generator Loss: 3.9977
Step 775/32 Epoch 1/1... Discriminator Loss: 0.8965... Generator Loss: 1.0004
Step 780/32 Epoch 1/1... Discriminator Loss: 0.4991... Generator Loss: 3.1379
Step 785/32 Epoch 1/1... Discriminator Loss: 1.3178... Generator Loss: 0.5694
Step 790/32 Epoch 1/1... Discriminator Loss: 1.1164... Generator Loss: 4.1838
Step 795/32 Epoch 1/1... Discriminator Loss: 2.1863... Generator Loss: 0.1887
Step 800/32 Epoch 1/1... Discriminator Loss: 1.0836... Generator Loss: 2.3164

Step 805/32 Epoch 1/1... Discriminator Loss: 1.2206... Generator Loss: 0.6515
Step 810/32 Epoch 1/1... Discriminator Loss: 0.8365... Generator Loss: 1.2401
Step 815/32 Epoch 1/1... Discriminator Loss: 1.2965... Generator Loss: 0.6689
Step 820/32 Epoch 1/1... Discriminator Loss: 0.6721... Generator Loss: 2.7134
Step 825/32 Epoch 1/1... Discriminator Loss: 1.4793... Generator Loss: 0.4511
Step 830/32 Epoch 1/1... Discriminator Loss: 1.0015... Generator Loss: 1.2636
Step 835/32 Epoch 1/1... Discriminator Loss: 1.4413... Generator Loss: 0.5978
Step 840/32 Epoch 1/1... Discriminator Loss: 0.9213... Generator Loss: 1.8787
Step 845/32 Epoch 1/1... Discriminator Loss: 1.2653... Generator Loss: 0.6151
Step 850/32 Epoch 1/1... Discriminator Loss: 0.7191... Generator Loss: 2.5549
Step 855/32 Epoch 1/1... Discriminator Loss: 0.8471... Generator Loss: 1.3320
Step 860/32 Epoch 1/1... Discriminator Loss: 0.7695... Generator Loss: 1.6645
Step 865/32 Epoch 1/1... Discriminator Loss: 0.8045... Generator Loss: 2.9469
Step 870/32 Epoch 1/1... Discriminator Loss: 1.5451... Generator Loss: 0.4160
Step 875/32 Epoch 1/1... Discriminator Loss: 1.7018... Generator Loss: 3.2734
Step 880/32 Epoch 1/1... Discriminator Loss: 1.6211... Generator Loss: 0.4434
Step 885/32 Epoch 1/1... Discriminator Loss: 0.7792... Generator Loss: 2.0877
Step 890/32 Epoch 1/1... Discriminator Loss: 2.2667... Generator Loss: 0.2495
Step 895/32 Epoch 1/1... Discriminator Loss: 1.1172... Generator Loss: 2.3819
Step 900/32 Epoch 1/1... Discriminator Loss: 1.5766... Generator Loss: 0.4031

Step 905/32 Epoch 1/1... Discriminator Loss: 0.8369... Generator Loss: 2.1257
Step 910/32 Epoch 1/1... Discriminator Loss: 1.2706... Generator Loss: 0.6180
Step 915/32 Epoch 1/1... Discriminator Loss: 0.7281... Generator Loss: 3.1564
Step 920/32 Epoch 1/1... Discriminator Loss: 0.9673... Generator Loss: 0.8385
Step 925/32 Epoch 1/1... Discriminator Loss: 1.0830... Generator Loss: 2.3553
Step 930/32 Epoch 1/1... Discriminator Loss: 0.9779... Generator Loss: 0.9623
Step 935/32 Epoch 1/1... Discriminator Loss: 0.9075... Generator Loss: 2.0458
Step 940/32 Epoch 1/1... Discriminator Loss: 0.9229... Generator Loss: 1.1468
Step 945/32 Epoch 1/1... Discriminator Loss: 0.8008... Generator Loss: 2.6714
Step 950/32 Epoch 1/1... Discriminator Loss: 1.3838... Generator Loss: 0.4683
Step 955/32 Epoch 1/1... Discriminator Loss: 1.2135... Generator Loss: 2.9762
Step 960/32 Epoch 1/1... Discriminator Loss: 1.4707... Generator Loss: 0.4935
Step 965/32 Epoch 1/1... Discriminator Loss: 0.9101... Generator Loss: 2.7595
Step 970/32 Epoch 1/1... Discriminator Loss: 1.3513... Generator Loss: 0.5167
Step 975/32 Epoch 1/1... Discriminator Loss: 0.5495... Generator Loss: 2.4992
Step 980/32 Epoch 1/1... Discriminator Loss: 1.7139... Generator Loss: 0.3681
Step 985/32 Epoch 1/1... Discriminator Loss: 0.9482... Generator Loss: 3.2569
Step 990/32 Epoch 1/1... Discriminator Loss: 1.6794... Generator Loss: 0.4227
Step 995/32 Epoch 1/1... Discriminator Loss: 1.1856... Generator Loss: 0.6899
Step 1000/32 Epoch 1/1... Discriminator Loss: 0.5812... Generator Loss: 3.3113

Submitting This Project¶

When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_face_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.